Research On The Mobile Agent-Based Data Fusion In Wireless Sensor Network

Wireless sensor network (WSN) is made of a large number of sensor nodes in the way of self-organization and multi-hop. The sensor nodes are limited in battery energy, computing ability and memory capacity, so it is an important issue that how to use the limited energy of sensor nodes in network design. We must make the monitoring range overlap each other, because the detection range and reliability of each node are limited, in order to enhance the accuracy and robustness of the network. So there is certain redundancy in the data of the sensor nodes. In the routing that each node transmits its detection data to the sink node, it is necessary for data fusion to reduce redundant information, save energy and prolong the network lifetime.This paper introduced WSN (Wireless Sensor Network) and the technology of data fusion. It got a conclusion of typical data fusion, analyzed the fusion on the layer of network, and proffered the mode of agent-based clustering data fusion and detail way of fruition.At first it proffered a division cluster algorithm based on the results of part fusion to construct the layer mode.This algorithm can control the cluster size by double cluster mode during the division.Based on the results of part fusion, it rejected error nodes before cluster forming for improving the accuracy. After clustering, the nodes in the region are divided into three layers:between the main cluster and sink node, we use client/server model; in the main cluster and the sub cluster, we use the mobile agent model. It optimizes the path of the mobile agent using PSO algorithm and gives fusion algorithm based on the optimal range of the detection data. The mobile agents carry the routing information and the node fusion algorithm and migrate to the nodes for data integration gradually. Through the experiments, the model can reduce the path loss, save energy, and then prolong the network lifetime.